|
|
|
|
|
|
|
|
|
|
| Rosemary Polsky-Newman, Vice President, ExerGen Biosciences |
|
| Rosemary Polsky-Newman, Ph.D. has over 20 years of experience in research and strategic planning in the bioinformatics, biotechnology, and medical-device industries. Prior to becoming Vice President of ExerGen Biosciences, she was the Director of Technology and Business Development for the New York Biotechnology Association. Formerly, she ran RPC Design Associates, a consulting firm dedicated to the development and design of immunoassays and nucleic acid tests. Earlier during her career, Dr. Polsky-Newman was Director of Corporate Planning at Medical and Scientific Designs following her tenure as a Senior Research Scientist at Baxter Dade Diagnostics. Dr. Polsky-Newman earned her Ph.D. from Tufts University and her B.A from Adelphi University.
|
|
The Utility of Aggregated Knowledge in the Life Sciences
Rosemary Polsky-Newman, Ph.D., Vice President, ExerGen Biosciences, Inc., 27 West 24th Street, Suite 401, New York, NY 10010, USA
Scientific fields in general, and the life sciences in particular, would benefit greatly from an automatically generated comprehensive database of extracted knowledge derived from scientific text. The utility of such a database is clear: comprehensiveness (vast size), consistency (reproducible extraction), automatically updated (always current). In addition, providing multiple databases across various related disciplines would facilitate cross-fertilization of ideas from disparate sources. However, the production of an automated system to extract knowledge from life science literature has been much more difficult than the production of generalized systems to extract meaningful knowledge from the web or newswires, primarily because of: the complexity of scientific language, the vast extent of the scientific dictionary, and the prevalence of synonyms and homonyms. Thus, the assembly of life sciences databases currently is accomplished by manual means in almost all cases. The GeneWays natural language processing platform, containing a full grammar as well as statistical and pattern matching components that can overcome the aforementioned problems will be described. The database(s) this system can populate have recall and precision approaching that of human curators, and have been used to perform research that would be otherwise impossible. The use of "Molecular Triangulation", on a prototype GeneWays database which contained over 125,000 full text articles, to identify candidate genes for Alzheimer's Disease will be described.
|
|
|
|
|
|
|
|
|
